Vision based automation system application (human language translator) (P-0203) (MFN 2822)

Abstract

This undergraduate dissertation describes the implementation of “Vision Based Automation System Application (Real Time Human Language Translator)”. My goal was to develop a system of human computer interaction which is able to detect and recognize the specific hand gestures made by a person standing at a distance from the camera or acquisition device. Previously, people have detected the hand gestures by focusing camera on the hands only but I am introducing an approach to take the image of whole scene and then separating hand region from the rest of the image. The application of this project is for disabled humans, who cannot speak well and use different hand gesture signs to tell or deliver their thoughts and feelings to others. This project is also useful in many places where sound or other means of communication is not possible, like factories where noise factor is very high and in space where sound waves can’t travel. In these cases, the best means of communication is vision based. The purposed algorithm in my project is able to detect and recognize the specific gestures made by a person, captured by a camera having either U.S.B. (Universal Serial Bus) or analog output. The image captured by the camera will be processed by a computing machine that is a P.C. Although many hand gesture recognition techniques are available but here I have used feature vector approach as mentioned in Shapiro & Stockman [1]. Moreover, the separation of hand, which is our region of interest, is also a challenging task. I have introduced variance based technique for this purpose. I have used MATLAB 7.9 for the implementation of my Algorithm, in which Image Processing Toolbox and Image Acquisition Toolbox are mainly used.